Austrocedrus from the Late Cretaceous of Antarctica and its implications for the early evolution of southern cypresses (Callitroideae, Cupressaceae)

Abstract

Premise

The modern distribution of the southern cypresses, Callitroideae (Cupressaceae), reflects a classic Gondwanan distribution. The initial diversification of Callitroideae is estimated to have occurred during the Cretaceous, yet fossils from this period are rare, hampering our understanding of its early evolutionary history.

Methods

A permineralized cluster of callitroid seed cones has been recovered from the Late Cretaceous Santa Marta Formation on James Ross Island, Antarctica. The specimen was sectioned using the acetate peel technique. We conducted a comparative morphological and anatomical investigation of the fossil with living members of Callitroideae and tested its phylogenetic placement using Bayesian and parsimony inference methods.

Results

Each cone has four opposite‐decussate ovuliferous complexes, a central columella, and subtending bifacially flattened shoots. This new fossil resembles living Austrocedrus in having ovuliferous complexes with a short, apically located bract tip and an abaxial row of vascular bundles. Phylogenetic reconstructions place the fossil sister to Austrocedrus chilensis, and we erect a new species, Austrocedrus vetus sp. nov.

Conclusions

Austrocedrus is known by a handful of Cenozoic fossils from Australia and a single surviving taxon, A. chilensis, in the Southern Andes of Chile and Argentina. Austrocedrus vetus is the oldest evidence of the genus, extending its stratigraphic record by 40 Ma. Our findings indicate that Austrocedrus diverged by the Campanian and experienced stasis in seed cone morphology. Furthermore, this fossil is the first occurrence of Austrocedrus in Antarctica, demonstrating that the genus was an element of Antarctic forest ecosystems during the Late Cretaceous.

The subfamily Callitroideae (in the cypress family, Cupressaceae) is a group of coniferous trees and woody shrubs restricted to the Southern Hemisphere, occupying a vast array of ecosystems from arid environments to rainforests (Hill and Brodribb, 1999). Callitroideae is composed of eight genera and ~35 species (Yang et al., 2022; Earle, 2025) that are distributed in Australasia (Callitris Ventenat, Libocedrus Endl., Diselma Hook f., Papuacedrus H. L. Li), southern Africa (Widdringtonia Endl.), and Patagonian South America (Austrocedrus Florin and Boutelje, Pilgerodendron Florin, Fitzroya Hook. f. ex Lindl.). The disjunct distribution of the Callitroideae in the Southern Hemisphere has made these conifers an important group for reconstructing the biogeographic history of the Gondwana continents (e.g., Mao et al., 2012; Kooyman et al., 2014; Crisp et al., 2019). Fossil evidence is an important source of data for uncovering the biogeographic history of these conifers, revealing historical distributions and diversity.

The Cenozoic fossil record of the subfamily is extensive, comprising shoots and cones of Callitris (Jordan, 1995; Paull et al., 2019), Libocedrus (Hill and Carpenter, 1989; Whang and Hill, 1999; Paull and Hill, 2009), Diselma (Paull and Hill, 2009), Austrocedrus (Hill and Carpenter, 1989; Paull and Hill, 2008), and Fitzroya (Hill and Paull, 2003; Paull and Hill, 2010) from Australia, and Papuacedrus (Zhou and Li, 1994; Wilf et al., 2009) from Patagonia and Antarctica. The fossil records of Papuacedrus, Fitzroya, and Austrocedrus suggest that these genera had a much wider distribution in the Southern Hemisphere in the past. For instance, Fitzroya and Austrocedrus have modern distributions in Patagonia, whereas fossils of these genera have been recovered from Paleogene deposits of Tasmania, Australia. Furthermore, Papuacedrus is currently restricted to New Guinea and the Moluccas, but was present in Patagonia during the Paleogene (Wilf et al., 2009). Some researchers have suggested that these biogeographic patterns indicate that these genera were likely broadly distributed across South America, Antarctica, New Zealand, and Australia during the final phases of the breakup of Gondwana (e.g., Wilf et al., 2009; Kooyman et al., 2014; Crisp et al., 2019).

Despite divergence time estimates that place the origin of Callitroideae in the Cretaceous, relatively few callitroid fossils are known from this period (Leslie et al., 2018; Crisp et al., 2019; Stull et al., 2021). A silicified seed cone was reported from the Upper Cretaceous Winton Formation (Tucker et al., 2013; Tucker, 2014) in Australia but has not been formally named (called “Callitris octothamna” in Peters, 1985; for a reassessment of some morphological features of this fossil, see Crisp et al., 2019: Methods S1). Wood fragments assigned to Cupressinoxylon hallei Kräusel, a proposed callitroid, have been found in mid‐Cretaceous deposits of Argentina (Brea et al., 2016) and Upper Cretaceous deposits of Antarctica (Santa Marta Formation; Pujana et al., 2017). Also from Argentina, wood from the Lower Cretaceous assigned to Austrocupressinoxylon barcinense Nunes, Bodnar et Escapa resembles some southern cypresses (Nunes et al., 2019).

The limited reports of Cretaceous Callitroideae alongside the distribution of living species of the subfamily suggest a once widespread Gondwanan range (Kooyman et al., 2014). This distribution implies the presence of Callitroideae in Antarctica, which has been documented by fossil wood (Cupressinoxylon hallei) from Cretaceous, Paleocene, and Eocene deposits (Pujana et al., 2014, 2015, 2017). Here, we describe a new fossil seed cone assignable to Callitroideae from the Upper Cretaceous (Campanian) Sitio Feliz locality of Antarctica. Morphological, anatomical, and phylogenetic studies reveal close affinities with Austrocedrus chilensis (D. Don) Pic. Serm. & Bizzarri within the Callitroideae.

MATERIALS AND METHODS

Fossil material

A single anatomically preserved vegetative shoot terminating in a cluster of seven seed cones was recovered in a calcium carbonate concretion from the Sitio Feliz locality of the Santa Marta Formation on James Ross Island of the Antarctic Peninsula. The Sitio Feliz locality (63°54′14″ S; 57°54′40″ W) hosts marine deposits belonging to the Beta Member of the Santa Marta Formation in the Marambio Group (James Ross Basin; Olivero et al., 1986; Scasso et al., 1991; Olivero, 1992, 2012a, 2012b; Milanese et al., 2020). The Beta Member is also synonymous with the upper section of the Lachman Crags Member (Crame et al., 1991; Pirrie et al., 1997); however, we have chosen to use the lithostratigraphic nomenclature provided in Olivero (2012a, 2012b; see also Milanese et al., 2020). The formation likely represents a progradational deltaic depositional system on a shallow marine shelf (Scasso et al., 1991; Olivero, 2012a) that received clastic sediments and terrestrial plants debris from neighboring continental areas of the Antarctic Peninsula and was not transported far distances (discussed in Scasso et al., 1991; Olivero et al., 2008). The Sitio Feliz deposits are dated to the middle Campanian (ca. 80 Ma) on the basis of magnetostratigraphic (basal section of Chron C33n; Milanese et al., 2020) and biostratigraphic (ammonite association 6; Olivero, 2012b) studies. Sitio Feliz plant fossils are represented by exceptionally preserved material of ferns, cycads, conifers, and angiosperms (Iglesias, 2016; Silva et al., 2025). The concretions were collected during the Antarctic summer of 2024 by U.S. Antarctic Program expedition G‐074E (led by coauthor B.A.A. and including coauthors S.Y.S. and A.I.), which was supported in part by the British Antarctic Survey.

The concretion containing the ovulate cone cluster was cut with a water‐cooled saw and sectioned using the cellulose acetate peel technique (Joy et al., 1956). The saw cut the cone cluster in half, revealing individual cones in various orientations on two separate faces (20327 Hbot and Itop). Both slabs broke into smaller pieces during peeling, and we gave each fragment a new identifier (Hbot, H1bot; Itop, I2top, I3top, I4top) and peeled only those fragments containing the seed cone cluster. A total of 1582 peels were made of the specimen: 844 peels of 20327 H/H1bot slab that had two cones in transverse section, and 738 peels of 20327 I/I2/I3/I4top that had five cones in various longitudinal to oblique section and their subtending shoot system. For light microscopy, peel sections were mounted on slides using Eukitt (O. Kindler, Freiber, Germany) mounting medium. Images were taken using a Nikon DS‐Fi3 camera on a Nikon Eclipse LV100ND microscope, a Nikon SMZ25 stereomicroscope with a Nikon DS‐Fi3 camera attachment, and a Canon EOS RP camera. Images and figures were edited for contrast and brightness and assembled into plates using Adobe Photoshop (Adobe Systems, San Jose, California, USA). All specimens and microscope slides are housed in the Division of Paleobotany Collection, Biodiversity Institute, University of Kansas (KUPB; specimen no. 20327; slide nos. 37569–37589, 31216–31233).

Serial images of three of the cones were aligned in Dragonfly (Comet Technologies Canada, Montréal, Canada). The alignments were segmented and three‐dimensionally reconstructed using Mimics 27.0 (Materialise, Leuven, Belgium). Videos of three‐dimensional reconstructions were rendered using Blender version 4.2 (Blender Development Team, Amsterdam, The Netherlands). Videos of peel series, reconstructions, and meshes can be found in Appendix S1. Image series and meshes of reconstructions are archived on MorphoSource (project ID: 000791849; Boyer et al., 2017).

Phylogenetic analysis

Phylogenetic relationships of the studied fossil, Austrocedrus vetus sp. nov., were evaluated with a combined molecular and morphological character matrix using Bayesian and parsimony inference methods.

Morphological character matrix

The morphological character matrix contains 83 characters and 21 extant Cupressaceae species and the new fossil (22 taxa total). Many of the characters were adopted from Andruchow‐Colombo et al. (2023) and Pfeiler et al. (2025) and several new characters were added. The taxon sampling included 15 species of Callitroideae (including the new fossil taxa) and seven species representing the other subfamilies of Cupressaceae. Morphological and anatomical data were sourced either by personal observation of material using histological sections and µCT scans (see below) or through descriptions previously published in the literature. In some cases, multiple species of a genus were sampled to capture the morphological variation within the genus (for the character matrix and character list with detailed descriptions, see Appendix S2). The character matrix consists of 51 seed cone, 28 vegetative, and 4 pollen and pollen cone characters (available on Morphobank P6278). The morphological character matrix contains 18.5% missing data and 5.2% inapplicable data.

The treatment of the morphological identity of the ovuliferous complexes of Cupressaceae has been debated and varies within the literature. Here, we follow the interpretation that both ovuliferous scale and bract are present and congenitally fused in mature seed cones (for a detailed review of this topic, see Matsunaga, 2025). Our interpretation differs from the original character construction for Cupressaceae by Matsunaga et al. (2021) and Andruchow‐Colombo et al. (2023). Therefore, we have rescored characters that describe bract and ovuliferous scale to reflect this interpretation.

In addition to published descriptions, observations of extant taxa were done from a combination of fresh material collected from the living conifer collection of California State Polytechnic University Humboldt (Arcata, California, USA), herbarium specimens from the U.S. National Museum of Natural History Herbarium (Washington, D.C.) and the Field Museum Herbarium (Chicago, Illinois, USA), and bioplastic embedded thin sections of woody seed cones provided by Ashley Kammet and Alexandru M.F. Tomescu at California State Polytechnic University Humboldt. Fresh materials that were collected and sectioned for anatomical study are housed in the anatomical collections of coauthor K.K.S.M. at the University of Kansas. Species that were collected for observation in this study, their respective collection sites and collection numbers, and identification numbers for herbarium material are listed in Appendix S3.

Fresh vegetative material was fixed in Formalin‐Aceto‐Alcohol and then dehydrated and embedded into paraffin wax using an ethanol‐Citrisolv‐paraffin graduated series (e.g., Ruzin, 1999). Specimens were sectioned at 10 µm thickness using an AO 820 rotary microtome (American Optical, Buffalo, New York, USA). When needed, specimens were soaked in a 5% fabric softener solution in a refrigerator to soften tissues (as detailed in El‐Abdallah et al., 2022). Sections were stained using either Walker's Sam quadruple stain protocol (1% Bismarck Brown in 50% ethanol, 1% Phloxine in 95% ethanol, Fast Green FCF‐Orange G in 1:1:1 ethanol–clove oil–methyl cellosolve solution) or a Safranin‐O Fast Green FCF staining protocol (1% Safranin O in 100% ethanol, Acidulated 70% ethanol, 0.05% Fast Green FCF in 100% ethanol (adopted from Moreno‐Sanz et al., 2020). Slides were mounted using Eukitt (O. Kindler, Freiber, Germany) mounting medium. Sections used to score character matrix were imaged (see Appendix S2).

Several seed cones, mostly from members of Callitroideae, were selected for µCT scanning to observe anatomy. μCT scans were performed on a Nikon XTH 225ST industrial CT system (Nikon Metrology, Brighton, Michigan, USA) located at the University of Michigan Earth and Environmental Sciences μCT facility (UM CTEES). A 1620 X‐ray detector panel (Perkin Elmer, Waltham, Massachusetts, USA) and a tungsten reflection target were used for each scan. Scan data and scan conditions for each specimen are archived on MorphoSource (project ID: 000818614). Observations of μCT scans for character scoring were done using Dragonfly.

Molecular data

We used the 18S and PHYP nuclear loci and matK and rbcL plastid loci for the molecular block of our data set. The sequences were obtained from GenBank (for accession numbers, see Appendix S3). We used the currently accepted species names as recognized by the Kew Plants of the World Online checklist (https://wcsp.science.kew.org/home.do). Sequences were aligned in MAFFT (Katoh and Standley, 2013) using default parameters. Alignment edges were manually trimmed and then all loci were concatenated using SequenceMatrix version 1.8 (Vaidya et al., 2011). The resulting molecular block contains 6117 characters and 21 taxa.

Analyses

The molecular and morphological data were combined in total evidence analyses in MrBayes (Huelsenbeck et al., 2001; Ronquist et al., 2012) and TNT version 1.6 (Goloboff et al., 2003, 2008; Goloboff and Catalano, 2016; Goloboff and Morales, 2023). The combined data set contains 6200 characters.

For the Bayesian analyses, the molecular data were separated into four partitions, one for each gene. We used Partition Finder 2 (Guindon et al., 2010; Lanfear et al., 2012, 2016) to evaluate the best model of evolution for each partition. Using the Akaike Information Criterion under a greedy search algorithm, a general time‐reversible model of rate substitution (GTR) was preferred for three loci with either gamma‐distributed rate variation among sites (+G; matK) or proportion of invariable sites (+I; PHYP, rbcl). The K80 model of evolution with gamma‐distributed rate variation among sites and proportion of invariable sites was preferred for 18S (K80 + I + G). The morphological characters were analyzed using the Mk model (Lewis, 2001) with gamma‐distributed rate variation across characters. Tree searches were performed using two parallel runs of the Metropolis‐coupled Markov chain Monte Carlo with three hot chains and one cold chain. Chain length was set to 100 million generations per run and they were sampled every 10,000 generations, resulting in 10,000 sampled trees. The standard deviation of split frequencies and acceptable effective sample size for parameters for each run was assessed in Tracer version 1.7.2 (Rambaut et al., 2018) to confirm convergence. The sampled distribution of trees was summarized using a 50% majority‐rule consensus tree with 25% of trees discarded as burn‐in. The input file and all output files for the Bayesian total evidence analysis is provided in Appendix S2.

Parsimony searches were initiated using the default parameters of xmult = , which include a round of random addition sequences (RAS) then refined by tree‐bisection‐reconnection (TBR) rearrangements, sectorial searches, and tree fusing. The search was set to stop after reaching the minimum length 100 times (xmult = hits100). The trees were subject to an additional round of branch swapping using TBR, and a strict consensus tree was calculated from the most parsimonious trees. A standard bootstrap analysis was performed using 1000 replicates. Alternative placements of the new fossil were accessed using maximum parsimony and default parameters of the PlaceMyFossil script in TNT (Catalano et al., 2025). Results of this analysis are provided in Appendix S2.

RESULTS

Systematics

Subclass —Pinidae (Gorozhankin, 1904)

Order —Cupressales (Link, 1829)

Family —Cupressaceae (Gray, 1822)

Tribe —Austrocedreae (Yang, 2022)

Genus —Austrocedrus Florin and Boutelje 1954

Species —Austrocedrus vetus Pfeiler, Matsunaga & Atkinson sp. nov.

Specific diagnosis —Shoot subtending cones with transition scale leaves. Vegetative shoots bifacially flattened with distinctly dimorphic leaves. Penultimate stem with parenchymatous pith surrounded by continuous cylinder of secondary xylem; cortex thin and fused with leaf bases. Lateral leaves strongly keeled, conduplicate; facial leaves bifacially flattened. Leaves supplied by singe terete vascular strand; single resin canal abaxial to vascular bundle. Ovulate cones 11 mm long, 5 mm in diameter bearing four opposite‐decussate ovuliferous complexes. Ovuliferous complexes dorsiventrally flattened, bearing a blunt, rounded apex; ultimate pair longer than penultimate pair; penultimate pair sometimes of unequal sizes, at least half the length of the ultimate pair. Bract tip small, blunt, located on apex of the ovuliferous complex. Both surfaces of ovuliferous complexes ridged, minutely papillate. Central columella ~1 mm long, hollow, tapered to sharp point, frequently with minute lobes at base. Cone axis short, parenchymatous pith surrounded by woody vascular strands lacking growth rings. Ovuliferous complexes supplied by a ring of vascular strands, vascular traces branch to form an abaxial and an adaxial row of bundles; bract bundle unbranched; vascular bundles with large amounts of secondary xylem. Three resin canals at base of ovuliferous complex, abaxial to vascular trace.

Holotype —KUPB 20327 Hbot, H1bot, Itop, I2top, I3top, I4top

Repository —Division of Paleobotany (KUPB), Biodiversity Institute, University of Kansas, Lawrence, Kansas

Locality —Sitio Feliz locality, James Ross Basin, Antarctic Peninsula

Stratigraphic and age —Santa Marta Formation, Beta Member, Marambio Group; Late Campanian (ca. 80 Ma; Olivero, 2012b; Milanese et al., 2020)

Etymology —Specific epithet vetus, meaning “of long standing” or “of a former time”

Description

Austrocedrus vetus is represented by seven anatomically preserved, mature, dehisced seed cones and a subtending bifacially flattened shoot system (Figures 1, 2, 3, 4, 5, 6; Appendix S1). All seven cones are likely to have been connected through the branching shoot system, but the connection was lost in the saw cut that revealed the cones. Three of the cones, however, remain in organic connection through a leafy branching shoot (Figures 2 and 5A–F; Appendix S1).

Figure 1.

Austrocedrus vetus sp. nov. seed cones. (A) Longitudinal section through a seed cone. KUPB 20327 H1bot no. 180. Scale bar = 1 mm. (B) Transverse section through a seed cone closer to the base of cone. Note three, dorsiventrally flattened ovuliferous complexes and the fourth separating from the central hollow columella. KUPB 20327 Itop no. 1. Scale bar = 1 mm. (C–E) Three‐dimensional reconstructions of three of the cones in the cluster. Top row: complete cone reconstruction. Each cone has two pairs of opposite‐decussate ovuliferous complexes, the ultimate pair (pink; ovuliferous complexes at left and right) longer than the penultimate pair (tan; ovuliferous complexes in front and back). Bottom row: with one ovuliferous complex removed to reveal central columella (white arrowheads). Black arrowhead indicates bract tips. Red arrowheads denote the highly ridged external surface of the ovuliferous complexes. (D, E) Cones with the penultimate pair of ovuliferous complexes (tan or complexes in front and back) of highly uneven length. KUPB 20327 Hbot and Itop.

Figure 2.

Austrocedrus vetus sp. nov. seed cones. (A, B) Five seed cones in various oblique‐longitudinal section. Each cone is highlighted a different color in panel B. Note leafy shoots subtending two cones at right (pink, orange). KUPB 20327 H1bot no. 650. Scale bar = 2 mm.

Figure 3.

Vascular geometry of ovuliferous complex of Austrocedrus vetus sp. nov. (A–E) Transverse sections of ovuliferous complexes with the traced vascular bundles (black) and the bract trace (orange). Sections are oriented with adaxial surface up and abaxial surface down. Metaxylem has been filled in to better indicate the orientation of each vascular bundle. (A) KUPB 20327 Itop no. 1. (B) KUPB 20327 I3top no. 300. (C) KUPB 20327 I4top no. 351. (D) KUPB 20327 I4top no. 445. (E) KUPB 20327 I4top no. 571. Scale bar = 500 µm.

Figure 4.

Anatomical details of Austrocedrus vetus sp. nov. lobed columella. (A) Transverse section through columella and base of ovuliferous complex. Note large central columella and smaller lateral columella lobe (red arrowhead); multilayered wall of columella composed of sclereids. Below columella are the vascular bundles that supply the base of an ultimate ovuliferous complex. At bottom of image, note resin canal that enters the base of the ovuliferous complex. KUPB 20327 Itop no. 1. Scale bar = 200 µm. (B) Tangential section of the columella with a lobe (red arrowhead). KUPB 20327 H1bot no. 600. Scale bar = 200 µm. (C) Radial section though the center of a seed cone with a small lobed columella (lobe at red arrowhead). Note vascular tissues entering opposite ovuliferous complexes (black arrowheads). KUPB 20327 Hbot no. 150. Scale bar = 200 µm.

Figure 5.

Anatomical details of Austrocedrus vetus sp. nov. seed cones. (A) Transverse section of base of cone (cone axis) showing central pith and vascular cylinder separated into six traces. KUPB 20327 Hbot no. 100. Scale bar = 200 µm. (B) Transverse section of base of cone (cone axis) showing central pith and vascular cylinder separated into six traces. KUPB 20327 H1bot no. 500. Scale bar = 100 µm. (C) Detail of secondary tissues of cone axis. KUPB 20327 Hbot no. 1. Scale bar = 100 µm. (D) Longitudinal section of secondary xylem with bordered pits (black arrowheads). KUPB 20327 H1bot no. 650. Scale bar = 50 µm. (E) Cross section at the base of ovuliferous complex showing separate vascular strands, and three abaxial resin canals that supply the base of the complex. KUPB 20327 H1bot no. 250. Scale bar = 200 µm. (F) Detail of epidermis of ovuliferous complex with short papillae. KUPB 20327 Hbot no. 100. Scale bar = 100 µm. (G) Detail of epidermis of cone axis. KUPB 20327 Hbot no. 100. Scale bar = 50 µm.

Figure 6.

Anatomical details of Austrocedrus vetus sp. nov. vegetative shoots. (A) Leafy shoots subtending the cones. Note the connection of the two shoots near their branching point. KUPB 20327 Hbot no. 100. Scale bar = 200 µm. (B) Detail of shoot in longitudinal section of A. Note tightly appressed scale leaves, and large amounts of secondary xylem. KUPB 20327 Hbot no. 100. Scale bar = 200 µm. (C) Uniseriate xylem ray of the wood of section in A–B. KUPB 20327 Hbot no. 100. Scale bar = 50 µm. (D) Three vegetative shoots near branching point. KUPB 20327 H1bot no. 300. Scale bar = 1 mm. (E) Transverse section of bifacially flattened shoot showing keeled lateral leaf (at right); at this level, the facial leaves are tightly appressed and indistinct. KUPB 20327 H1bot no. 600. Scale bar = 200 µm. (F) Transverse section of bifacially flattened shoot with strongly keeled lateral leaf (left black arrowhead) and lateral leaf tip (right black arrowhead). Note distinct, bifacially flattened facial leaf (red arrowhead). KUPB 20327 H1bot no. 450. Scale bar = 500 µm. (G) Longitudinal section through leaf showing vascular bundle. KUPB 20327 H1bot no. 400. Scale bar = 200 µm. (H) Detail of lateral leaf with hypodermis directly underneath epidermis, a thin vascular bundle (black arrowhead), and a resin canal directly abaxial to the vascular bundle. KUPB 20327 H1bot no. 450. Scale bar = 50 µm. (I) Transverse section of keeled lateral leaf. KUPB 20327 H1bot no. 450. Scale bar = 100 µm.

Ovulate cones

The ovulate cones are oblong in shape, 11 mm in length and 5 mm in width. Each cone consists of four valvate ovuliferous complexes in two opposite pairs (Figure 1A–E). The penultimate pair of ovuliferous complexes are much smaller than the ultimate pair, measuring 6.0–7.5 mm long (Figure 1C, D). In at least two of the cones the penultimate pair of ovuliferous complexes are of slightly unequal lengths and are at least half the length of the ultimate pair (Figure 1D, E). The ultimate pair of ovuliferous complexes range from 8.5 mm to 11 mm long.

All ovuliferous complexes are 5 mm at the widest point and have a blunt, rounded apex (Figure 1B–E). The ovuliferous complexes are dorsiventrally flattened, ~1 mm at thickest, and contain highly longitudinally ridged surfaces (Figure 1). The bract tip is small, blunt, and located near the apex of the ovuliferous complex (Figure 1A, C–E) and does not extend beyond the tip of the ovuliferous scale. In many of the cones the bract tip is often level with the outer surface of the ovuliferous complex, appearing as a small mound of tissue (Figure 1C–E). In one cone, the penultimate ovuliferous complexes contain horizontally oriented grooves or ridges in the upper portion of the complex as revealed in the 3D reconstruction (Figure 1C). This may be a bract tip that is broader than the bract tip seen in the ultimate pair.

Each of the cones contains a central columella that extends ~1 mm from the cone axis (Figure 1). The columellae are hollow at the base and do not contain vascular tissue (Figures 1A, B and 4A). The hollow cavity disappears as the columella continues and tapers to a sharp point. The wall of the columella is composed of several layers of sclereids and parenchyma cells (Figures 1B and 4A). Some of the cones contain a smaller cavity adjacent to the prominent, central columella (Figure 4A–C). These smaller cavities, which are lobes of the columella, vary significantly in size, measuring 85–230 µm tall and 315–560 µm wide. The columella lobes have blunt apices and are often not clearly visible in the three‐dimensional reconstructions due to their small size (Figure 1C–E). Some of the columella cavities contain a detached ring of dark cells that may be an epithelial layer (Figure 4A).

The cones are located terminally on leafy shoots (Figure 5A, B). The cone axes are short due to the immediate divergence of the ovuliferous complexes, as in all Callitroideae seed cones. The pith of the cone axis is 130–170 µm wide and is made up of thin‐walled parenchyma cells (Figure 5A, B). The vascular tissue of the cone axis is composed almost entirely of secondary xylem (≤ 360 µm thick) and does not contain growth rings. Outside of the xylem cylinder is a very thin layer of secondary phloem and a thin cortex with small resin canals that lead to the base of the ovuliferous complexes. The xylem cylinder of the stem separates into approximately six strands at the base of each cone (Figure 5A, B). The epidermis at the base of the cones is 10–25 µm thick and does not appear to have a cuticle preserved (Figure 5A, G).

Each ovuliferous complex is supplied by a vascular cylinder of separate strands (i.e., no continuous vascular cambium) containing a central pith (Figure 3A; 4A, B; and 5E). The abaxial collateral bundle, which has the xylem above the phloem, corresponds to the bract bundle. The bract bundle extends for most of the length of the ovuliferous complex and is unbranched (abaxial orange bundle in Figure 3). The adaxial bundle of the cylinder divides laterally in a single plane to form part of an adaxial row of bundles with phloem adaxial to xylem. The lateral bundles on either side of the cylinder divide laterally and vertically, contributing to the adaxial row of bundles on the upper side of the ovuliferous complex. In addition, the vertical divisions produce six to eight abaxial bundles. The bundles in the abaxial row are oriented with xylem adaxial to phloem. The vertical branching also produces a few bundles on the outer edges of the ovuliferous complex that have phloem pointing toward the lateral edges of the ovuliferous complex. In all, this branching pattern creates a dorsiventrally flattened ring of vascular strands with protoxylem poles pointed toward the center of the ovuliferous complex. In distal regions of the ovuliferous complex, the abaxial bundles terminate, leaving only the adaxial row of bundles that extends to the apex of the ovuliferous complex.

The vascular bundles in the ovuliferous complexes are composed mostly of secondary xylem (Figures 1B and 3). The tracheids of the secondary xylem contain bordered pits (Figure 5D), though these are rarely preserved. The secondary xylem tracheids have thick secondary cell walls. The ground tissue in the ovuliferous complexes is composed of large sclereids and parenchyma cells. Three resin canals supply the base of all the ovuliferous complex (Figure 5E) and do not extend beyond half the length of that complex. Several thin and short canals originate de novo, randomly throughout the ovuliferous complexes; most, however, are oriented near the abaxial surface (Figure 3C). The resin canals can reach a diameter of 200 µm, but most are thinner. The outer surface of the ovuliferous complex is covered in papillae ~55 µm high (Figure 5F). The tissues of all the ovuliferous complex contain large amounts of hyphae and other potential fungal remains (Appendix S1). The cones appear woody and mature, yet no seeds or ovules are preserved inside them, which suggests they were fully dehisced prior to fossilization. Further, the cone cluster was likely transported during diagenesis, as evidenced by the lack of a well‐preserved cuticle. However, the cones do not appear to be heavily abraded, given the preservation of epidermis and epidermal papillae.

Vegetative shoots

The shoots at the base of the cones are ~1 mm long and covered in scale‐like transition leaves (Figure 6A, B). These transition leaves are small, 0.35–0.40 mm long. The penultimate shoot system comprises stems that are oval in transverse section and contain decussate, highly appressed, dimorphic lateral and facial scale leaves (Figure 6D–I). The pith of the stem is 475 × 51 µm and is composed of large parenchyma cells, but most axes lack a preserved pith. The stem contains a 150–500 µm thick layer of secondary xylem that lacks growth rings. Xylem rays are uniseriate and 3–10 cells tall (Figure 6C). The cortex of the stem is very thin, crushed between secondary vascular tissues and tightly appressed, mostly fused, scale‐leaves (Figure 6D–F).

The leaves of the bifacially flattened shoots are dimorphic, with strongly keeled, conduplicate lateral leaves and small facial leaves that are appressed to the stem (Figure 6E, F). The lateral leaves measure 280 µm thick and clasp the stem (Figure 6F). The facial leaves measure 810 µm wide and 215 µm thick. A single vascular trace supplies facial and lateral leaves, 30 µm in diameter, and is subtended by a resin canal ~50 µm in diameter (Figure 6G–I). The leaves have a uniseriate to biseriate hypodermis (Figure 6H).

Phylogenetic analyses

Austrocedrus vetus is consistently placed sister to A. chilensis in both parsimony and Bayesian analyses (Figure 7). Its position has relatively high support from both analyses: posterior probability of 0.88 in the Bayesian analysis and a bootstrap value of 85%. The parsimony analyses resulted in two most parsimonious trees, reflecting uncertainty in relationships within Libocedrus. However, A. vetus is placed in the same position in both most parsimonious trees, which is supported by four synapomorphies: (1) separate strands of secondary xylem of the cone axis (character 19); (2) the bract tip located near the apex of the ovuliferous complex (character 25); (3) separate vascular strands entering the base of the ovuliferous complex (character 34); and (4) the absence of a prominent resin canal supplying the base of the ovuliferous complex (character 47). In the PlaceMyFossil analysis, the optimal placement of A. vetus is on the A. chilensis branch, with suboptimal placements requiring at least four extra steps (see Appendix S2; there are seven best suboptimal placements within the Austrocedrus, Pilgerodendron, Libocedrus, Fitzroya, and Diselma grade).

Figure 7.

Phylogenetic relationships of Austrocedrus vetus sp. nov. based on Bayesian total evidence and parsimony analyses. The fossil species is indicated by a dagger (†) and bold font. Branch lengths represent those of the Bayesian analysis. Support values are labeled for each node: posterior probabilities (top) and bootstrap values (bottom). Only support values below 0.95 and 95 are shown.

DISCUSSION

Comparisons

Morphological comparisons to living genera of Callitroideae

Valvate seed cones with few, opposite‐decussate or whorled ovuliferous complexes are produced by members of Callitroideae. Austrocedrus vetus is most like Papuacedrus, Libocedrus, Austrocedrus, and Pilgerodendron, which have cones with four, valvate, opposite‐decussate, dorsiventrally flattened ovuliferous complexes (Figure 8). Many of these genera produce cones in dense clusters, like A. vetus. Apart from Papuacedrus, which does not appear to have a columella, the other aforementioned genera have seed cones with elongate columellae (Farjon, 2005; Paull, 2006). Additionally, most of these genera and A. vetus have bifacially flattened shoots with distinctly dimorphic leaves. The only exception is Pilgerodendron (Figure 8C), which has shoots with monomorphic leaves and is therefore easily distinguished from A. vetus.

Figure 8.

Shoots and seed cones of four members of Callitroideae. (A) Austrocedrus chilensis. Abaxial surface of shoot. (B) Libocedrus plumosa. Adaxial surface of shoot. (C) Pilgerodendron uviferum. (D) Papuacedrus papuana USNM 3436404. Adaxial surface of shoot.

Seed cones of Libocedrus and Pilgerodendron have very long bract tips that diverge in the middle region of the ovuliferous complex and terminate in an acute, elongated spine (Figure 8B, C). Some species, such as Libocedrus chevalieri J. Buchholz and L. plumosa (D. Don) Druce, have an elongate bract that emerges from the upper half of the ovuliferous complex (Hill and Carpenter, 1989). Papuacedrus has a small bract projection (2–3 mm) that is located on the lower half of the ovuliferous complex (Figure 8D; Paull, 2006). The ovuliferous complexes and the bract tips of Papuacedrus are often outwardly reflexed at dehiscence (Paull, 2006). Austrocedrus is unique among this group in having ovuliferous complexes with a minute blunt bract tip at the apex of the ovuliferous complex (Figure 8A). In Austrocedrus, the length of the separated bract tip varies from a short mound of tissue to a small projection (Figure 8A; Paull, 2006). The position and size of the bract tip in A. vetus is most similar to bract tips of A. chilensis and varies only in that the bract tip of A. vetus is fused to the ovuliferous scale for the entire length of the ovuliferous scale and does not extend to a small projection. The penultimate pair of ovuliferous complexes in A. chilensis are about one‐third to half the length of the ultimate pair (Figure 8A; Paull and Hill, 2008). In A. vetus the penultimate pair can be of unequal lengths and are at least half the length of the ultimate pair (Figure 1D, E).

It is also worth noting that Diselma, Widdringtonia (within Callitroideae‐Cupressaceae), and Tetraclinis Masters (Cupressoideae‐Cupressaceae) possess seed cones with four opposite decussate ovuliferous complexes. The cones of Diselma are small and inconspicuous (~3 mm in length), and the columella reaches or extends beyond the ovuliferous complexes in mature cones (Farjon, 2005; Dörken and Jagel, 2017). Widdringtonia seed cones contain ovuliferous complexes that are equal in size and thick and rhomboidal in cross section. The columella of Widdringtonia is often deeply lobed and has a blunt apex. In contrast, A. vetus has cones with a columella that is barely lobed and has a sharp apex. These traits clearly indicate that Diselma and Widdringtonia differ from A. vetus. Tetraclinis, once considered to be within Callitroideae (Li, 1953) but now placed in the subfamily Cupressoideae (Saxton, 1913; Brunsfeld et al., 1994; Gadek et al., 2000; Jagel and Stützel, 2003), contains valvate seed cones with rhomboidal ovuliferous complexes. The pairs of ovuliferous complexes are similar in length, the penultimate pair is wider than the ultimate pair, and the cones lack a columella. Seed cone anatomy of Tetraclinis differs from the new fossil in numerous ways, but primarily in having a very short bract bundle that supplies the bract tip (Lemoine‐Sebastian, 1969; K. C. Pfeiler et al., personal observation). Further, Tetraclinis does not have bifacially flattened vegetative shoots and can therefore be excluded from further comparison with A. vetus.

As mentioned, some of the cones of A. vetus have penultimate ovuliferous complex pairs that are of uneven lengths (Figure 1D, E). The asymmetry of cones that results from uneven ovuliferous complex length has been discussed by Crisp et al. (2019: Methods S1 and discussion of fossil Libocedrus mesibovii Hill et Carpenter). These authors suggest, through their observations of images and specimens, that extant Pilgerodendron is the only member of the subfamily that exhibits uneven lengths of the penultimate pair of ovuliferous complexes. Our observations of extant A. chilensis demonstrate that this species can occasionally produce asymmetric cones, generally found among cones that are tightly clustered together. Rather than being a unique feature of Pilgerodendron (and some fossils), we suggest that ovuliferous complex asymmetry is simply the result of differential growth of ovuliferous complexes after pollination. This could happen because of direct contact of cones in tightly packed clusters or some other environmental feature like orientation of the cluster‐bearing branch. Whether this feature is a discrete, evolutionarily inherited character that has been lost in some genera (i.e., Libocedrus), as posited by Crisp et al. (2019), remains unclear. Nevertheless, the presence of both symmetric and asymmetric cones in extant Austrocedrus, Pilgerodendron, and A. vetus suggests that this feature can vary within a single plant and may be more common in modern genera of Callitroideae than is formally documented.

Anatomical comparisons to living genera

The similarity of vascular anatomy in the ultimate pair of ovuliferous complexes of A. chilensis and A. vetus is striking. Austrocedrus chilensis contains a cylinder of separate vascular strands, each composed mostly of secondary xylem that branches at the base of the complex. The most abaxial bract bundle extends through most of the length of the ovuliferous complex unbranched. The remaining vascular tissue branches to produce an adaxial row of bundles and an abaxial row of “inverted bundles” (Figure 9A). This pattern was first recognized by Lemoine‐Sebastian (1969), who noted that A. chilensis differs from Libocedrus in having a few abaxial bundles that flank the central bract bundle. Pilgerodendron uviferum (D. Don) Florin and Libocedrus plumosa lack the abaxial bundles and have only a row of adaxial bundles and a single bract bundle (Figure 9B, C). The new fossil contains an abaxial row of bundles that resembles A. chilensis (Figure 3). Lemoine‐Sebastian (1969) also notes that the penultimate pair of ovuliferous complexes in Libocedrus and Austrocedrus have fewer number of vascular traces than the ultimate pair, which is corroborated by our observations of Austrocedrus, P. uviferum, and L. plumosa (Figure 9). Libocedrus plumosa (Figure 9B) and, to a more extreme degree, P. uviferum (Figure 9C) contain numerous resin canals in the ovuliferous complexes. Austrocedrus chilensis (Figure 9A) and A. vetus have very few resin canals (Figure 3).

Figure 9.

Transverse digital sections of mature, dehisced seed cone of three Callitroideae species from µCT scans. All three species have valvate, opposite‐decussate ovuliferous complexes and a central columella. Below, ovuliferous complexes with vascular traces highlighted (not to scale). Blue = vascular bundles; orange = bract bundle. (A) Austrocedrus chilensis; note adaxial row of bundles and an abaxial row of bundles that flank the bract bundle; see also the hollow columella. (B) Libocedrus plumosa; note adaxial row of bundles and hollow columella. (C) Pilgerodendron uviferum; note adaxial row of bundles, abundance of resin canals, and hollow columella. Scale bar = 1 mm.

In many members of Cupressaceae, and especially those of Callitroideae, the bract is supplied by a single trace (oriented with xylem above phloem) that is the first diverging bundle from the vascular cylinder that supplies the ovuliferous complex (Lemoine‐Sebastian, 1969, 1972). In Austrocedrus, this bundle remains unbranched and extends toward the apex of the ovuliferous complex suppling the free bract tip (see µCT scan; Lemoine‐Sebastian, 1969). Anatomical studies of Papuacedrus seed cones do not exist; however, it is likely that it exhibits a similar pattern to Libocedrus (see µCT scan; Lemoine‐Sebastian, 1969) and Pilgerodendron (see µCT scan) with a single, unbranched bundle that supplies the bract tip on the lower half of the ovuliferous complex. The bract bundle in A. vetus extends for over half the length of the scale (Figure 3); this suggests, from anatomical evidence alone, that the bract tip is positioned on the apex of the ovuliferous complex, as seen in A. chilensis.

The columella of A. vetus is hollow, probably once resin filled, as in the three observed living species in this study (Figure 9). A study by Dörken and Jagel (2017) investigated the histological components of columellae in several species of Callitroideae in pollination‐stage cones. Dörken and Jagel's study documents the columella of A. chilensis and P. uviferum as lacking intercellular spaces and resin canals and as being composed of isodiametric parenchyma cells. In Diselma, Widdringtonia, and Callitris the columella is hollow, lined with epithelial cells, and formed schizogenously (as is typical in resin canal formation; Beck, 2005; Dörken and Jagel, 2017). Our results reveal that in mature cones of A. chilensis, P. uviferum, L. plumosa, and A. vetus the columella is hollow at maturity. This cavity, therefore, likely develops schizogenously, later in development (post‐pollination), as it does in Diselma, Widdringtonia, and Callitris.

Deeply lobed columellae, a common feature of Widdringtonia and Callitris, are not described in A. chilensis (Dörken and Jagel, 2017). The lobes of the columella in A. vetus are structurally different from those in Widdringtonia and Callitris because they are not readily visible externally (Figure 1C–E) and are observable only through peel sections (Figure 4). It is possible that, like A. vetus, extant A. chilensis exhibits rare lobing that is not externally visible and is therefore not previously described. Alternatively, this feature may have once been present in the genus (as evidenced by A. vetus) and has been lost in the extant species. Further investigation of fully mature seed cones of A. chilensis is needed to determine whether lobing of the columella is present.

Comparisons to fossil Callitroideae

Eight fossil cones have been assigned to Callitroideae (Berry, 1919; Peters, 1985; Hill and Carpenter, 1989; McIver, 2001; Hill and Paull, 2003; Wilf et al., 2009; Paull and Hill, 2009, 2010), six of which have four ovuliferous complexes and are compared here. Austrocedrus tasmanica Hill et Carpenter, from the early Miocene of Tasmania, has four opposite‐decussate ovuliferous complexes that differ in size, and each contains a bract tip at the apex of the complex (some authors call this a tubercle; Hill and Carpenter, 1989). Both A. vetus and A. tasmanica have ovulate cones in which the penultimate pair of ovuliferous complexes is 75% of the length of the ultimate pair, which is longer than that of the extant A. chilensis. Austrocedrus tasmanica has scale leaves that cover the bases of the ultimate pair of ovuliferous complexes, which is absent in A. chilensis and A. vetus. Austrocedrus tasmanica is much smaller than A. vetus, measuring 4 mm long × 1 mm wide.

An ovulate cone described from the Early Oligocene of Tasmania was found in association with foliage assigned to A. australis Paul et Hill and was informally referred to as the “Lea River ovulate cone” (Paull and Hill, 2008). This cone is similar to modern Austrocedrus, but the bracts on the penultimate pair of ovuliferous complexes are broader and flap‐like. This trait differs from A. vetus, and beyond this, comparison with the Lea River ovulate cone is limited due to preservation.

A small, possibly pollination‐stage seed cone attached to foliage assigned to L. leaensis Paull et Hill is described from the Early Oligocene of Tasmania (Paull and Hill, 2009). The cone is early in development and has immature ovuliferous complex morphology. The cone has four opposite‐decussate ovuliferous complexes with an ultimate pair that is longer than the penultimate pair. Each ovuliferous complex has an elongate bract projection, which the authors liken to those in extant Libocedrus. This fossil is clearly different from A. vetus in having elongated bract tips, even if early in development. Further comparisons are not possible due to major morphological changes that occur during maturation of cupressaceous seed cones.

Libocedrus mesibovii Hill et Carpenter is known from a single ovulate cone from the upper Oligocene deposits of Tasmania (Hill and Carpenter, 1989; see Crisp et al., 2019: Methods S1). The penultimate pair of ovuliferous complexes are of drastically unequal size in relation to one another. The upper pair of ovuliferous complexes, however, are similar to one another in size. The authors note that the asymmetry of ovuliferous complexes is not typical of any other genus in the subfamily. This feature has been identified in other genera within this group (Pilgerodendron; Crisp et al., 2019) and in Austrocedrus (as mentioned above). Libocedrus mesibovii has a scar where the bract tip has supposedly been abraded off. Crisp et al. (2019) suggested that this fossil is likely Pilgerodendron due to the uneven length of ovuliferous complexes of the penultimate pair. However, this character could be more common than is formally described and may not be a feature that should be used for taxonomic assignment. Given the abraded bract tip projection of the ovuliferous complexes and a lack of other defining characters, we cannot further compare this fossil to A. vetus. Nonetheless, the taxa are similar in having uneven lower ovuliferous complexes and a bract tip located at the apex of the complex.

Papuacedrus prechilensis (Berry) Wilf et al., a compression fossil from the early Eocene of South America, is represented by well‐preserved vegetative shoots and a terminal seed cone (Wilf et al., 2009). The seed cone contains a small flap‐like bract that terminates at the central region of the ovuliferous complex, as seen in P. papuana. Austrocedrus vetus is clearly different from P. prechilensis in the position of the bract tip at the apex of the ovuliferous complex.

Widdringtonia americana McIver is a compression fossil from the Tuscaloosa Formation of North America (Berry, 1919; McIver, 2001). The placement of this fossil in Widdringtonia has been questioned and is based on morphological features that are also found in other members of Cupressaceae. Personal observation of the fossil and discussions by other authors (Stockey et al., 2005; Leslie et al., 2012: supplemental data) indicate that it lacks diagnostic features of Widdringtonia and that its placement within the highly derived and entirely Southern Hemisphere genus may not be warranted. Widdringtonia americana lacks any further features that can be compared to A. vetus, due to its preservation.

The new fossil is not assignable to any previously described fossils within Callitroideae. While the phylogenetic relationships of A. vetus and the fossils discussed above are not provided in the present study, these relationships are worth pursuing in detail and will be addressed in a forthcoming study. The exceptional preservation of the fossil allows us to use both anatomical and morphological characters to determine the phylogenetic relationships of this new specimen. Austrocedrus vetus is remarkably similar to extant A. chilensis, containing salient features (e.g., bifacially flattened shoots; valvate seed cones with four, opposite‐decussate, dorsiventrally flattened ovuliferous complexes; apically positioned bract tips; a central columella; presence of an abaxial row of vascular bundles in the ovuliferous complex) that support its inclusion in Austrocedrus. Austrocedrus vetus differs from A. chilensis in having a lobed columella rather than an unlobed columella. Furthermore, A. vetus has cones with a penultimate pair of ovuliferous complexes that are longer than those of A. chilensis. These differences indicate that A. vetus and A. chilensis are distinct species.

Antiquity of Austrocedrus

Today, Austrocedrus is a monotypic genus endemic to Southern Andes forests in Chile and Argentina. The fossil record of Austrocedrus is known from three fossils: A. australis from the lower Oligocene deposits of Tasmania (Paull and Hill, 2008); A. tasmanica from the lower Miocene deposits of Tasmania (Hill and Carpenter, 1989); and one report of shoots assigned to Austrocedrus sp. from the Miocene deposits of Chile (Troncoso and Romero, 1993).

Divergence‐time estimates for Austrocedrus vary in several recent node‐calibrated dating analyses (Mao et al., 2012; Yang et al., 2012; Leslie et al., 2018; Crisp et al., 2019; Sudianto et al., 2020; Stull et al., 2021; Liu et al., 2022). Further complicating matters, the relationships within some genera of Callitroideae—Austrocedrus, Papuacedrus, and Libocedrus + Pilgerodendron—are not well resolved. The topology presented in the present study, Papuacedrus and Austrocedrus as successive sister lineages to the rest of the subfamily, is reported in the majority of studies (Leslie et al., 2012; Mao et al., 2012; Yang et al., 2012; Lu et al., 2014; Leslie et al., 2018). However, more recent studies show Papuacedrus sister to Austrocedrus, which together form a clade with Pilgerodendron + Libocedrus (Stull et al., 2021; Liu et al., 2022). Regardless of topology, these studies generally infer an Early or Late Cretaceous origin of Austrocedrus (Mao et al., 2012; Leslie et al., 2018; Liu et al., 2022). Austrocedrus vetus broadly corroborates the molecular divergence estimates and extends the stratigraphic record of the genus by ~40 million yr. Because this fossil can be confidently placed in Austrocedrus, it will prove to be a significant fossil calibration for divergence‐time analyses.

There have been several attempts to reconstruct ancestral areas for the Callitroideae, yet the analyses seem to be highly sensitive to fossil selection, model choice, and topology used in the analysis (Mao et al., 2012; Yang et al., 2012; Crisp et al., 2019; Liu et al., 2022). Liu et al. (2022) considered Australasia as the ancestral range for the genus when fossils are included in their analyses. Mao et al. (2012) suggested South America as the ancestral range for the node of Austrocedrus and the rest of Callitroideae. When considering the Oligocene and Miocene Australian distribution of Austrocedrus, Crisp et al. (2019) posited that the genus was vicariant in South America and Australia. In addition to having differing topological positions of Austrocedrus, these analyses are greatly hindered by a scant fossil record of Callitroideae as a whole, rendering ancestral area reconstructions unclear. Now, considering the addition of A. vetus, the fossil record of Austrocedrus reveals that the lineage was present in Antarctica as early as the Campanian, in Australia as late as the early Miocene (A. tasmanica), and in South America during the Miocene. The fossil record of Austrocedrus alongside its current distribution in South America reveals that the genus was widespread across Gondwana and experienced extirpation in Antarctica and Australasia. Although A. vetus cannot provide a clear indication of the geographic origins of Callitroideae, perhaps its inclusion in future biogeographic analyses could help elucidate the origins of the group.

Austrocedrus in the Antarctic flora

Paleo‐Antarctic rainforest lineages (PARLs) are modern rainforest lineages that were once part of an ancient Gondwanan flora that extended across South America, Australia, and Zealandia when those continents remained connected or in close proximity to Antarctica (Kooyman et al., 2014). Today, most PARLs exist in present‐day Oceania. Austrocedrus, on the other hand, represents one of the more rare “western survival” taxa, which survived in modern‐day South America while going extinct in Australasia (Kooyman et al., 2014). However, until now there has been no direct evidence for Austrocedrus fossils as part of the Gondwanan flora prior to the final breakup of the southern supercontinent. For the first time, we have macrofossil data to demonstrate that Austrocedrus was well established in Antarctica during the Late Cretaceous and exhibited morphology and anatomy resembling its modern relative, providing evidence of morphological stasis in the genus. Austrocedrus vetus also provides information on which Cupressaceae taxa, and conifers in general, were present in the western Antarctic flora. Most of what is known about the Late Cretaceous flora of Antarctica is from palynological records (e.g., Dettmann and Thompson, 1987; Bowman et al., 2014; Barreda et al., 2019; Santiago et al., 2024), yet pollen of Cupressaceae notoriously lacks features that are diagnostic below the family level (e.g. Bortenschlager, 1990; Kurmann, 1994; Tedeschini et al., 2021). The same holds true for the petrified wood fragments known in these deposits (Pujana et al., 2014, 2015, 2017). This makes it difficult to understand which genera of Cupressaceae contributed to the Late Cretaceous flora of Antarctica. Thus, our results underscore the importance of meso‐ and macro‐fossils of reproductive organs in furthering the understanding of Antarctic vegetation in the Cretaceous.

Austrocedrus chilensis is restricted to southern South America and tolerates a wide array of environmental conditions, ranging from arid regions at high altitudes to extremely wet forests at low altitudes (Veblen and Schlegel, 1982; Dodd et al., 1998; Hill and Brodribb, 1999; Dodd and Afzal‐Rafii, 2000; Gyenge et al., 2005; Earle, 2025). It also occurs in mesic Nothofagus Blume mixed forests (Veblen and Lorenz, 1987; Hill and Brodribb, 1999). In the more recent past, the cooler and wetter climate of the middle to late Holocene allowed the expansion of Austrocedrus/Nothofagus forests (Whitlock et al., 2006). Given the highly diverse habitat preference of Austrocedrus in today's flora, we refrain from using A. vetus as an indicator for a specific paleoclimate for the Sitio Feliz fossil Antarctic flora. As we learn more about the community composition and other paleoenvironmental elements of Late Cretaceous Antarctica, we will clarify how these plants survived and adapted to an ever‐changing Gondwanan landscape.

CONCLUSIONS

Very little is known of the fossil record of Callitroideae in Gondwana, and even less is known about the anatomical traits that these conifers possessed during the Cretaceous. Austrocedrus vetus is the oldest known fossil of the genus, extending its stratigraphic record back by ~40 million yr, and adds to the knowledge of extinct species diversity in the fossil record. It is also the only occurrence of Austrocedrus in Antarctica, confirming its once broader distribution in Gondwana before extirpation in Australasia and Antarctica. Austrocedrus vetus is anatomically and morphologically similar to extant A. chilensis, revealing that by the Late Cretaceous Austrocedrus had diverged and experienced morphological stasis of seed cone morphology. Further, these findings support a Cretaceous origin for Austrocedrus as estimated in molecular divergence time analyses for the callitroid clade. The anatomical preservation of plants in the Sitio Feliz flora provides extraordinary data for comparative morphological studies, giving us insights into the evolution of the conifers that once flourished on Gondwana.

AUTHOR CONTRIBUTIONS

K.C.P.: Conceptualization, formal analysis, funding acquisition, investigation, project administration, visualization, writing—original draft. B.A.A.: Conceptualization, data curation, funding acquisition, supervision, validation, writing—review and editing. K.K.S.M.: Funding acquisition, supervision, validation, data curation, writing—review and editing. S.S.: Funding acquisition, validation, supervision, writing—review and editing. A.I.: writing—review & editing. A.A.‐C.: Data curation, writing—review and editing.

Supporting information

Appendix S1: Animations and meshes of 3D reconstructions of the fossil specimens as well as a plate illustrating fungi present in the specimen.

Appendix S2: Input and output files for total evidence phylogenetic analyses. Aligned molecular sequences are provided as FASTA format files. The morphological character matrix is provided as a nexus file formatted for Mesquite. Also included is a list of character definitions and explanations for the morphological character matrix. Input files and output files for MrBayes and TNT analyses are included. We provide a plate containing histological cross‐sections of shoots in select Callitroideae that were used for scoring morphological characters.

Appendix S3: GenBank accession numbers for DNA sequence data and herbarium or collection numbers for specimens used for comparative morphology (histological or µCT scans).

DATA AVAILABILITY STATEMENT

Data are available on MorphoSource (fossil data project ID: 000791849; seed cone µCT scans of living species project ID: 000818614) and on Morphobank (P6278).